CMAJ OPEN Research

The current and future financial burden of hospital admissions for heart failure in : a cost analysis

Dat T. Tran MPH, Arto Ohinmaa PhD, Nguyen X. Thanh MD PhD, Jonathan G. Howlett MD, Justin A. Ezekowitz MBBCh MSc, Finlay A. McAlister PhD, Padma Kaul PhD

Abstract

Background: Heart failure is a costly health condition and a major public health concern. We sought to examine the costs of hospital admissions for heart failure between fiscal years 2004 and 2013 in Canada and to model the future costs to 2030. Methods: Canadian Institutes for Health Information Discharge Abstract Database was used to identify admissions to hospital with heart failure as the primary diagnosis between fiscal years 2004 and 2013. Multiple linear regression models were used to calculate the trend in prevalence and extrapolate these to 2030. Canadian Institutes for Health Information patient cost estimates were used to identify costs of hospital admissions for heart failure. Generalized linear models were used to estimate average annual costs per heart failure patient. We conducted a sensitivity analysis including all admissions for heart failure in any diagnostic field. Results: In 2013, 45 600 (95% confidence interval [CI]: 43 800–47 200) patients were admitted with heart failure as the primary diag- nosis, accounting for $482 (95% CI $464–$500) million. By 2030, we estimate 54 000 (95% CI 49 000–60 000) patients and costs of $722 (95% CI $650–$801) million, with older adults (age ≥ 80 yr) accounting for 52% of costs. Including admissions for which heart failure was a secondary diagnosis increases the total cost to $2.8 (95% CI $2.6–$3.0) billion in 2030. Interpretation: As in other developed countries, hospital costs related to heart failure in Canada are on the rise. Older adults are the main consumers of such hospital services. Strategies to improve outpatient care to reduce rates of admission for heart failure are needed.

eart failure is a costly health condition and a major Methods public health problem. It is estimated that 2%–3% H of the population in developed countries has heart Annual volume of patients from 2004 to 2013 (all of failure, and the prevalence increases to 8% among patients Canada except ) aged more than 75 years.1 Heart failure is the single most The Canadian Institute for Health Information Hospital common reason for hospital admission.2,3 In the United Discharge Abstract Database from 2004 to 2013 for all States, it was projected that there would be 8.5 million people Canadian provinces and territories except Quebec was used (3% of the US population) with heart failure in 2030, which to identify hospital admissions with a primary diagnosis of would cost the US health system $53 billion.4 heart failure (International Statistical Classification of Diseases It has been hypothesized that a combination of and Related Health Problems, 10th revision, code I50). improved survival in heart failure patients5–7 and population Canadian population estimates for the same period from aging8 is expected to increase the burden of this condition Statistics Canada9 (minus Quebec) were used as denominators in Canada. However, little is known as to the current and to calculate annual prevalence rates per 100 000 population by future financial burden of heart failure on the Canadian sex and age (< 60 yr, 60–69 yr, 70–79 yr and ≥ 80 yr). Patients health care system. with multiple admissions for heart failure during the same The objectives of our study were to examine hospital year were counted once. admission costs for heart failure between fiscal years 2004 and 2013 in Canada and, based on these costs, model the future prevalence and admission costs to 2030. Although Competing interests: None declared. long-term heart failure prevalence and costs in Canada may This article has been peer reviewed. change as a result of new therapies or changes in manage- ment strategies, forecasts based on current trends can serve Correspondence to: Padma Kaul, [email protected] as useful benchmarks to examine the effects of future inno- CMAJ Open 2016. DOI:10.9778/cmajo.20150130 vations on health care costs.

© 2016 Joule Inc. or its licensors CMAJ OPEN, 4(3) E365 CMAJ OPEN Research

Estimating national volume of heart failure patients used Canadian population estimates (medium population from 2004 to 2013 (including Quebec) growth — M5) from 2014 to 20308,17 to calculate the expected We did not have data on admissions for heart failure for the number of heart failure patients for each year. The annual province of Quebec. To calculate national estimates for the total admission costs for 2014–2030 were then derived by burden of heart failure, we assumed the prevalence rates of multiplying the expected number of admitted patients with hospital admission for heart failure in Quebec to be the aver- the annual admission cost per patient. age across all other provinces and territories and used multiple linear regression (natural logarithm of prevalence rate as the Sensitivity analyses dependent variable) with fiscal year, sex, age group and the We conducted 3 sensitivity analyses. First, we examined the interaction between sex and age as independent variables in the uncertainty around our annual cost estimates by varying the model. Joinpoint regression was used5,10–12 to detect significant estimated average annual admission cost per patient within its changes in annual prevalence rates between 2004 and 2013. 95% confidence interval (CI). Second, we added low and high We recorded significant turning points at anα level of 0.05 to population growth scenarios8,17 into projections of prevalence use as a knot in the multiple linear regression model. Model and costs from 2014 to 2030. Finally, we recalculated preva- assumptions (homoscedasticity, normality of residuals, and the lence and cost estimates based on admissions for which heart linearity of relationships between the outcome and fiscal year failure was coded as either a primary or secondary diagnosis. as a continuous predictor) were checked and explored for Our primary analysis is based on admission for which heart unusual and influential observations by examining residuals failure was coded as the primary diagnosis. However, heart and leverage values. The number of patients was derived by failure has been shown to be associated with other conditions, multiplying the estimated prevalence rate from the multiple such as acute coronary syndromes and renal disease. It is linear regression model with the Canadian population (includ- therefore possible that when heart frailure is present as a sec- ing Quebec) for each age and sex group for each year. ondary diagnosis, it could contribute to hospital costs.

Annual hospital costs per heart failure patient from Results 2004 to 2013 The cost of each admission to hospital was obtained from the Characteristics of the study population (all provinces Canadian Institute for Health Information, where the patient except Quebec) cost estimates were made using an algorithm that took case mix Between 2004 and 2013, there were 421 121 hospital admis- groups, resource intensity weight, cost per weighted case and sions with a primary diagnosis of heart failure in Canada, length of stay into consideration. This calculation provides aver- excluding Quebec (Table 1). The mean expected length of stay age costs incurred through the direct care of hospital inpatients for an admission increased from 7.5 days in 2004 to 8.3 days in (e.g., inpatient nursing, diagnostic and therapeutic costs).13,14 For 2013. However, the median acute length of stay remained sta- case mix groups that did not have cost available in the Canadian ble at 6 days. The number of admissions and patients both Institute for Health Information data, we used the Case reached the lowest point in 2007 and then increased. Female Costing Initiative15 instead. All costs were converted to 2014 patients outnumbered male patients in the first 3 years. Patient dollars using the Bank of Canada inflation calculator.16 Annual mean age increased over time, as did the proportion of patients inpatient costs for a heart failure patient were derived by sum- over the age of 80 years. Diabetes, chronic obstructive pulmo- ming costs of all hospital admissions of that patient in a year. nary and renal diseases were the most common comorbidities We fitted a generalized linear model with gamma distribution among heart failure patients. The largest increase in comor- and log link to estimate average annual admission costs per bidity rate was for diabetes, which increased from 29.3% in patient for the entire study period (2004–2013). The variables 2004 to 45.7% in 2013. Proportion of patients with more than included in the model were fiscal year, patient sex and age 1 admission with heart failure as the primary diagnosis per year group. Other patient factors were not included because they increased from 17.9% in 2004 to 18.8% in 2013. were already incorporated into the Canadian Institute for Health Information patient cost estimate algorithm. National volume of heart failure patients from 2004 The annual total hospital admission costs associated with to 2013 (including Quebec) heart failure in Canada from 2004 to 2013 were derived by The results of the multiple linear regression model to estimate multiplying the estimated number of patients admitted with the prevalence rate of heart failure hospital admissions over the annual hospital cost per patient. time for each age and sex group are shown in Appendix 1 (available at www.cmajopen.ca/content/4/3/E365/suppl/DC1).​ Projection of heart failure prevalence and costs from These rates declined over time in all groups (Figure 1). The 2014 to 2030 prevalence rates were more than 100 times higher in the oldest We assumed that the trend seen during 2004 to 2013 would compared with the youngest groups in both sexes. Hospital continue. The coefficients from the multiple linear regression admissions for heart failure were more prevalent in men in all and generalized linear models described earlier were used to age groups. predict the expected prevalence rate and admission cost per In contrast to the declining trend in hospital admission heart failure patient for each year from 2014 to 2030. We prevalence rates, the absolute number of heart failure patients

E366 CMAJ OPEN, 4(3) CMAJ OPEN Research

Table 1: Characteristics of heart failure patients admitted to hospital, 2004–2013 (excluding Quebec)

Characteristic 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Admissions to 43 102 42 039 41 033 40 390 40 899 41 084 41 585 42 298 43 163 45 528 hospital, no. Mean ELOS*, d 7. 5 7. 5 7. 7 7. 8 7. 9 8.0 8.1 8.3 8.4 8.3 (SD) (4.1) (4.1) (4.4) (4.7) (4.8) (5.0) (5.1) (5.1) (5.3) (5.3) Median aLOS (IQR) 6 6 6 6 6 6 6 6 6 6 (3–11) (3–11) (3–11) (3–11) (3–11) (3–11) (3–11) (3–11) (3–11) (3–11) Patients 34 311 33 600 32 876 32 542 33 004 32 995 33 144 33 753 34 365 36 142 Male/female ratio 0.96 0.98 0.98 1.00 1.02 1. 01 1. 01 1. 01 1.03 1.04 Mean age, yr (SD) 76.8 76.9 77.1 77.1 77.2 77.1 77.6 77.6 77.8 77.9 (12.1) (12.4) (12.3) (12.5) (12.5) (12.7) (12.4) (12.5) (12.6) (12.7)

Patient proportion by age group, yr, %† < 60 8.8 9.2 8.8 9.0 9.2 9.0 8.4 8.6 8.5 8.6 60–69 13.5 13.7 13.5 13.8 13.7 14.5 14.0 13.8 13.7 14.0 70–79 29.8 28.7 27.8 27.1 27.0 26.2 25.7 25.1 24.6 23.8 ≥ 80 47.8 48.4 50.0 50.1 50.2 50.3 51.9 52.5 53.2 53.6 Patient comorbidity, % AMI 11. 7 11. 7 12.0 11. 9 12.7 11. 6 11. 6 11. 1 10.5 9.6 COPD 19.5 19.4 20.1 19.7 20.5 20.2 21.0 21.1 20.7 20.2 Dementia 5.2 5.1 5.4 5.2 5.5 5.3 5.9 6.0 5.9 5.9 Diabetes 29.3 30.4 40.0 40.9 40.6 45.0 44.7 44.9 45.5 45.7 Renal 19.0 20.0 18.6 18.0 18.8 14.6 14.4 14.2 13.9 14.3 Patients with > 1 1 7. 9 1 7. 7 1 7. 8 1 7. 4 1 7. 4 1 7. 6 18.2 18.4 18.6 18.8 admission, %

Note: aLOS = acute length of stay, AMI = acute myocardial infarction, COPD = chronic obstructive pulmonary diseases, ELOS = expected length of stay, IQR = interquartile range, SD = standard deviation. *Number of days in hospital assigned to each admission based on CIHI grouping methodology. †Total percentage may not be exactly 100% because of rounding

Female HF patients Male HF patients 2500 2500

2000 2000

1500 1500

1000 1000

500 500 Prevalence rate per 100 000 Prevalence rate per 100 000 0 0 2004 2007 2010 2013 2004 2007 2010 2013

Observed rate Fitted: ≤ 60 yr Fitted: 60–69 yr Fitted: 70–79 yr Fitted: ≥ 80 yr

Figure 1: Prevalence rates of hospital admissions for heart failure (HF) in Canada, 2004–2013.

CMAJ OPEN, 4(3) E367 CMAJ OPEN Research admitted to hospital increased over time. There were 34 311 Sensitivity analyses patients admitted with heart failure in 2004, and this num- Varying annual admission cost per patient within its 95% CI ber increased to 36 142 in 2013 (Table 1). Extrapolating to led to a maximum deviation of 1% of total costs in 2013. The include Quebec, we estimated 44 400 (95% CI 42 500– deviation increased to 5.8% in 2030. This translated to a 46 300) patients were admitted in 2004, and 45 600 (95% range of total admission costs from $477 to $487 million and CI 43 800–47 200) in 2013, with a primary diagnosis of from $681 to $764 million in 2013 and 2030, respectively. heart failure. Compared with medium population growth, low popula- tion growth resulted in 1500 fewer projected heart failure Annual admission costs per patient from 2004 to 2013 patients and high population growth resulted in an additional Results of the generalized linear model estimating annual 2500 projected patients in 2030. Consequently, the low popu- admission cost per heart failure patient are presented in lation growth scenario would be associated with a $22 million Appendix 1. Admission costs increased by 1.4% annually and reduction and the high population growth scenario would be were higher for men and younger patients. The average per associated with an additional $33 million in admission costs. patient costs in 2004 and 2013 were estimated at $9700 and Results of prevalence and cost estimates based on admis- $11 000, respectively (Table 2). sions with heart failure as either the primary or secondary diag- Total admission costs from 2004 to 2013 increased gradu- nosis are provided in Appendix 3 (available at www.cmajopen​ ally (Figure 2). Using 2014 dollar values, the total cost was .ca/content/4/3/E365/suppl/DC1).​ Extending the analyses to $415 (95% CI $397–$432) million in 2004 and $482 (95% CI include admissions with heart failure as a secondary diagnosis $464–$500) million in 2013, equivalent to a 1.7% annual more than tripled (n = 175 201 [95% CI 164 185–186 838]) increase during this period. In 2013, older adults (age ≥ 80 yr) the estimate of patients admitted to hospital with heart failure accounted for 48.1% of costs; patients aged 70–79 years, in 2030. Consequently, total costs associated with heart fail- 60–69 years and less than 60 years accounted for 27%, 16.2% ure were projected to be $2.82 (95% CI $2.65–$3.01) billion and 8.7% of costs, respectively. in 2030.

Projection of prevalence and costs from 2014 to 2030 Interpretation Results of cost projection to 2030 are presented in Appendix 2 (available at www.cmajopen.ca/content/4/3/E365/suppl​/DC1). Patients who were admitted to hospital with a primary diag- We estimated about 48 000 (95% CI 45 000–51 000) patients nosis of heart failure cost the Canadian health care system in 2020 and about 54 000 patients (95% CI 49 000–60 000 $415 million in 2004 and $482 million in 2013. These admis- patients) in 2030. Annual admission costs per patient were sions therefore accounted for 0.8% of hospital spending in projected to increase to about $12 000 in 2020 and about 2013.18 We estimate that the cost of admissions with heart $14 000 in 2030. Total admission costs were projected to failure as the primary diagnosis will increase substantially by increase to $722 (95% CI $650–$801) million in 2030 (Figure 2030 to about $720 million (50% increase from 2013), 2). This reflects a 49.8% increase from 2013 to 2030. The assuming the current patterns continue. Older adults (age older adults (age ≥ 80 yr) were projected to account for 52% ≥ 80 yr) are the largest consumers of hospital services associ- of total costs in 2030. ated with heart failure. Including admissions where heart fail-

Table 2: Annual hospital admission cost for patients admitted to hospital for heart failure (in 2014 dollar values)

Characteristic 2004 2007 2010 2013 Mean cost per patient, $ 9 679 10 092 10 522 10 970 Women, yr All ages 9 543 9 950 10 374 10 816 < 60 10 044 10 472 10 918 11 383 60–69 9 917 10 339 10 780 11 239 70–79 9 498 9 903 10 325 10 765 ≥ 80 8 714 9 085 9 472 9 876 Men, yr All ages 9 816 10 234 10 670 11 125 < 60 10 331 10 771 11 230 11 709 60–69 10 200 10 635 11 088 11 560 70–79 9 770 10 186 10 620 11 073 ≥ 80 8 963 9 345 9 743 10 158

E368 CMAJ OPEN, 4(3) CMAJ OPEN Research ure is a secondary diagnosis increased the cost estimate to Furthermore, we used more recent data (of 10 consecutive $2.8 billion by 2030. years up to 2013) and thus were able to take into account the Inpatient prevalence rates are declining in both sexes. This slightly decreasing trends in admission prevalence in all age might be due to improved management of heart failure in groups, whereas the previous study by Johansen and col- outpatient clinics.19 However, the absolute number of patients leagues used only a single data point of fiscal year 1996/97.20 admitted to hospital is increasing. This is largely due to the We projected that a patient admitted to hospital with a pri- change in population demographics. Statistics Canada has mary diagnosis of heart failure would cost the health care sys- projected marked increases in the oldest age groups where tem about $14 000 per year in 2030. This is higher than the heart failure is most prevalent: the proportion of people aged average cost of $8200 in the United States (converted to 2014 80 years and over is expected to increase by 73% in the next Canadian dollars assuming admission accounts for 80% of 15 years, from 1.1 million people (4.1% of total population) in medical costs4 and 80% of patients with heart failure are admit- 2015 to 1.9 million (6.1% of total population) in 2030.8 Our ted21). This could be explained by our use of case mix groups, estimate of the increasing trend in the absolute number of which take comorbidities into consideration, whereas the US heart failure patients admitted to hospital is more conservative study modelled medical costs attributable to heart failure only. than that reported previously.20 This is probably because we Furthermore, patterns of care may also account for the differ- focused on hospital admission with a primary diagnosis of ence in cost estimates between the 2 countries. We have previ- heart failure. Extending the analysis to include admissions ously shown that heart failure patients in Canada have lengths where heart failure was coded as a secondary diagnosis more of stay that are longer than their US counterparts.22 than tripled our estimates of the number of patients admitted. Limitations We assumed that the trend in hospital admission patterns seen during 2004 to 2013 would continue. Long-term heart failure s 60 prevalence and costs in Canada may change as a result of new therapies or changes in management strategies. It should be noted that the continuous widening of 95% CIs of the esti- 45 mates (Figure 2) suggests increased uncertainty over time. We did not have hospital admissions data from Quebec and assumed prevalence rates in that province to be the mean across all other provinces and territories. However, we exam- 30 ined variation across provinces and found no statistically sig- nificant difference in prevalence rates between provinces where we have data. This result strengthened our assumption. 15 Finally, this study only included hospital costs and does not include other costs, such as physician, outpatient and drug costs, which are of increasing importance as the management 0 of heart failure shifts from the inpatient to the outpatient set- Estimated no. of HF patients, thousand 2005 2010 2015 2020 2025 2030 ting.19 However, about 80% of total medical costs associated with heart failure can be attributed to hospital costs.4,23 800 Conclusion

s As with other developed countries, the hospital resources 600 related to the care of heart failure patients in Canada are increasing. Hospital admissions for which heart failure is the primary diagnosis cost the country $482 million in 2013, and these costs are projected to increase to $720 million by 2030. 400 Including costs where heart failure is the secondary diagnosis increased the total cost in our estimates to a sobering $2.8 billion by 2030. Older adults are the main consumers of hos- 200 pital services associated with heart failure. Strategies to

Estimated cost, $ million improve outpatient care to reduce admission rates are needed, as are innovations that provide seamless, cost-effec- 0 tive and evidence-based­ interventions from the emergency 2005 2010 2015 2020 2025 2030 department through to discharge. Although long-term pro- jection is subject to assumptions and uncertainties, we believe Figure 2: Estimated admission costs for heart failure and number of that our study offers a reasonable indication of future heart failure patients, 2004–2013, and projected for 2014–2030 (in resource needs for heart failure in inpatient settings from 2014 $ values). Error bars incidate 95% confidence intervals. today’s perspective.

CMAJ OPEN, 4(3) E369 CMAJ OPEN Research

References 15. OCCI Costing Analysis Tool. : Ontario Case Costing Initiative; 2013. Available: https://hsimi.ca/occp/occpreports/ (accessed 2015 Feb. 20). 1. Bui AL, Horwich TB, Fonarow GC. Epidemiology and risk profile of heart 16. Inflation Calculator. : Bank of Canada; 2015. Available: www.bankof​ failure. Nat Rev Cardiol 2011;8:30-41. canada.ca/rates/related/inflation-calculator/?page_moved=1 (accessed 2015 2. Neumann T, Biermann J, Erbel R, et al. Heart failure: the commonest reason May 15). for hospital admission in Germany: medical and economic perspectives. Dtsch 17. Population projections for Canada (2013 to 2063), provinces and territories Arztebl Int 2009;106:269-75. (2013 to 2038): technical report on, ethodology and assumptions. Ottawa: Sta- 3. Ross H, Howlett J, Arnold JM, et al. Treating the right patient at the right tistics Canada; 2014. Available: www.statcan.gc.ca/pub/91-620-x/91​-620​ time: access to heart failure care. Can J Cardiol 2006;22:749-54. -x2014001-eng.htm (accessed 2015 Feb. 28). 4. Heidenreich PA, Albert NM, Allen LA, et al.; American Heart Association 18. National health expenditure trends, 1975 to 2014. Ottawa: Canadian Institute Advocacy Coordinating Committee; Council on Arteriosclerosis, Thrombosis for Health Information; 2014. and Vascular Biology; Council on Cardiovascular Radiology and Intervention; 19. Ezekowitz JA, Kaul P, Bakal JA, et al. Trends in heart failure care: has the Council on Clinical Cardiology; Council on Epidemiology and Prevention; incident diagnosis of heart failure shifted from the hospital to the emergency Stroke Council. Forecasting the impact of heart failure in the United States: a department and outpatient clinics? Eur J Heart Fail 2011;13:142-7. policy statement from the American Heart Association. Circ Heart Fail 2013;​ 20. Johansen H, Strauss B, Arnold JM, et al. On the rise: the current and pro- 6:606-19. jected future burden of congestive heart failure hospitalization in Canada. 5. Jhund PS, Macintyre K, Simpson CR, et al. Long-term trends in first hospi- Can J Cardiol 2003;19:430-5. talization for heart failure and subsequent survival between 1986 and 2003: a 21. Unroe KT, Greiner MA, Hernandez AF, et al. Resource use in the last 6 population study of 5.1 million people. Circulation 2009;119:515-23. months of life among medicare beneficiaries with heart failure, 2000-2007. 6. Yeung DF, Boom NK, Guo H, et al. Trends in the incidence and outcomes Arch Intern Med 2011;171:196-203. of heart failure in Ontario, Canada: 1997 to 2007. CMAJ 2012;184:E765-73. 22. Kaul P, Reed SD, Hernandez AF, et al. Differences in treatment, outcomes, 7. Tu JV, Nardi L, Fang J, et al. National trends in rates of death and hospital and quality of life among patients with heart failure in Canada and the admissions related to acute myocardial infarction, heart failure and stroke, United States. JACC Heart Fail 2013;1:523-30. 1994-2004. CMAJ 2009;180:E118-25. 23. Dunlay SM, Shah ND, Shi Q, et al. Lifetime costs of medical care after heart 8. Table 052-0005: Projected population, by projection scenario, age and sex, failure diagnosis. Circ Cardiovasc Qual Outcomes 2011;4:68-75. as of July 1, Canada, provinces and territories. CANSIM (database). Ottawa: Statistics Canada; 2014. Available: www5.statcan.gc.ca/cansim/pick-choisir​ ?lang=eng&p2=33&id=0520005 (accessed 2015 Feb. 28). Affiliations: School of Public Health (Tran, Ohinmaa, Thanh), Univer- 9. Table 051-0001: Estimates of population, by age group and sex for July 1, Can- sity of ; Institute of Health Economics (Ohinmaa, Thanh), ada, provinces and territories. CANSIM (database). Ottawa: Statistics Canada; ; Department of Medicine (Howlett), University of Calgary, 2014. Available: www5.statcan.gc.ca/cansim/pick-choisir?lang=eng&p2=33 Calgary; Department of Medicine (Ezekowitz, McAlister, Kaul), Univer- &id=0510001 (accessed 2015 Feb. 28). sity of Alberta; Canadian VIGOUR Centre (Ezekowitz, McAlister, Kaul), 10. Surveillance Research Program, Statistical Methodology and Applications University of Alberta, Edmonton, Alta. Branch, Joinpoint Regression Program Online Help System Version 4.1.1. Bethesda (MD): National Cancer Institute; 2014. Contributors: Dat Tran and Padma Kaul conceptualized the study. Arto 11. Pham TM, Fujino Y, Kubo T, et al. Premature mortality due to stroke and Ohinmaa and Thanh Nguyen provided input into costing methodology. trend in stroke mortality in Japan (1980-2005). Eur J Public Health 2011;​ Dat Tran conducted the analyses and drafted the manuscript. Padma 21:609-12. 12. Kim HJ, Fay MB, Feuer EJ, et al. Permutation tests for joinpoint regression Kaul made significant revisions to the manuscript. All of the authors criti- with applications to cancer rates. Stat Med 2000;19:335-51. cally reviewed the manuscript for intellectual content. Dat Tran had full 13. Patient Cost Estimator. Ottawa: Canadian Institute for Health Information; access to the data and serves as the guarantor of the work. 2015. Available: www.cihi.ca/pce (accessed 2015 Feb. 26). 14. Cost of a standard hospital stay. Ottawa: Canadian Institute for Health Infor- Supplemental information: For reviewer comments and the original mation; 2015. Available: http://indicatorlibrary.cihi.ca/display/HSPIL/Cost+​ submission of this manuscript, please see www.cmajopen.ca/content/4/3​ of+a+Standard+Hospital+Stay (accessed 2015 Feb. 26). /E365/suppl/DC1

E370 CMAJ OPEN, 4(3)